The admin that turns income into a trade

Running this as a business comes down to four habits: records kept as jobs happen, tax reserved at the point of payment, scope written before work starts, and a monthly hour on your own numbers. Those habits, far more than assessment quality, separate the earners still here in year two from those who stopped in month seven.

This is the whole operating picture at summary depth, with the post that covers each part in detail linked from its section. Nothing here is tax, legal or financial advice, and every rule mentioned differs by country - the general shape is transferable, the specifics are not, and checking yours is not optional.

Money in: what actually reaches you

Start from take-home rather than headline price, because four separate deductions sit between a buyer's payment and your account and only one is usually advertised.

The commission is the visible one. Processing charges are the flat one, and flat charges do not scale down, which is why the smallest job on a card is the one most likely to be underwater. Currency conversion is the invisible one when buyer, platform and bank are not in the same currency, and it is frequently taken twice. Withdrawal minimums are the fourth, and they are not a fee at all - they are a permanent slice of your money you cannot spend, which is why withdrawal thresholds and trapped float is worth reading before choosing where to work.

The consequence for platform selection is direct: compare platforms on take-home, not commission, because a generous cut behind a high withdrawal minimum pays a beginner nothing for two months.

Timing is a separate problem from amount. Money that arrives six weeks after the work does not pay this month's bills however large it is, and payout schedules and cash flow is the part that catches people who priced correctly and still ran short.

Records, kept at the moment or not at all

Any record-keeping process longer than the job's own admin will be abandoned by week three. That is the entire design constraint, and record keeping that takes one minute is built around it.

What to capture, per job: date, buyer reference, format, agreed scope, fee, deductions, net received, and delivery date. How long to keep it is set by your tax authority rather than by convenience; the IRS's recordkeeping guidance, for one, says records must be kept "as long as needed to prove the income or deductions on a tax return". That is one row. After thirty rows it will tell you which format to stop offering, which is the whole argument of tracking take-home per format and something no amount of intuition produces.

The reason to keep them always arrives months after the moment you could have, which is why records you will wish you kept is the post to read in week one rather than month nine.

Two adjacent habits do most of the remaining work. Separating personal and work money into a distinct account removes the worst afternoon this trade will ever give you, which is reconstructing a year of mixed transactions from memory. And keeping receipts that count matters because reconstructing what a payment was for, eleven months later, is the part everybody underestimates.

Tax, as an orientation rather than an instruction

Every jurisdiction differs and this post guesses about none of them. What is general is the shape.

Income from this work is income, whatever the platform calls it. Tips and gifts are the easiest kind to forget precisely because they are untracked, and tips and gifts as income is a short read that prevents a specific unpleasant surprise.

Money deducted before it reached you was still earned by you, and it is separately a cost - failing to record it understates your costs and overstates nothing, which is the point of what counts as a cost of earning.

Reserve as you go rather than at a deadline. An annual reckoning against money already spent is the single most common self-inflicted wound in self-employment, and setting money aside for tax as you go is the mechanism that removes it entirely. Lateness carries its own price, separate from the tax: in the UK, HMRC charges an initial £100 penalty for a late Self Assessment return, then daily penalties of £10 after three months, plus interest on tax paid late.

Two threshold questions arrive later and are worth knowing exist: whether and when you must register, covered in registering as self-employed, and whether sales tax or VAT ever touches you, which for most earners it does not - sales tax and VAT, in passing explains the exceptions rather than the rule.

There is a point at which a single conversation with a professional costs less than one mistake, and when to get an accountant is a more useful frame than a revenue figure.

Scope, which is where disputes are prevented

Nearly every dispute in this trade has the same two ingredients: an unclear scope and a vague delivery.

The prevention is two sentences sent before work starts, saying what will be delivered and by when. It costs twenty seconds and is your entire case in the one job out of fifty that goes wrong - the scope sentence that prevents most disputes is the shortest high-value habit on this site.

The written version belongs in the listing too, because a scope written as prose is a scope you cannot cite: writing a scope box buyers actually read covers the format.

Revisions are the leak that scope does not automatically close. An unnamed revision policy means the answer is always yes, which means the answer is always free, and handling a revision request sets the boundary before it is tested.

When something does go wrong, whoever can produce the agreed brief and the delivery record wins, largely regardless of who was right - evidence that actually wins a dispute is the inventory of what counts. And knowing when not to fight matters as much: the hours and the review usually cost more than the fee, which is the calculation in when to concede a dispute. Getting paid: holds, chargebacks and walking away covers the money side of the same situations, and it is deliberately not repeated here.

Buyers, incidentally, do not experience any of this as adversarial. The commissioning side's account of how a job is expected to open, run and close reads as a description of ordinary professional conduct, and an earner who scopes in writing looks like a professional to them rather than like someone being careful.

Capacity, which is the real ceiling

Every job needs you personally, so there is a hard ceiling and it is calculable. Finding your own income ceiling is the arithmetic, and knowing the number is more useful than aspiring past it.

The consequence is that growth here comes from price and format rather than volume beyond a point, which is what what scaling actually means here works through. The one genuine exception is the part of the work that can be detached from you - a template, a guide, a recorded explainer - covered in productising part of the work.

Before the ceiling, capacity is mostly a scheduling problem. Demand clusters in exactly the hours most people want back, which is why this trade quietly expands to fill an entire week unless something stops it: working hours that do not eat your week is that something. Batching work into sessions recovers the setup time that five separate evenings waste, and a weekly rhythm that holds is the structural version.

Burnout in a per-job trade is specific rather than generic. When nothing is measured in hours, nothing tells you how many you worked until it is already a problem, which is the case made in burnout in a per-job trade and the reason the monthly review below matters operationally rather than just financially.

Risk you do not control

Suspension does not have to be deserved to be total, and appeal processes are slow everywhere. Account risk, and not betting everything on one platform is the general treatment, and the correct time to read it is before anything happens.

Two mitigations are cheap. Read the platform rules in advance rather than after, since suspension for an honest mistake is indistinguishable from suspension for a dishonest one - read the platform rules before, not after. And accept that everything persuasive about your listing belongs to the platform hosting it, which is what makes a reputation that survives a platform a deliberate project rather than a by-product.

Fee and rule changes are the quieter version of the same exposure. They arrive by email, usually with notice, and usually in one direction - fee changes, and reading the terms email is a habit rather than a strategy.

You also hold other people's material, which is an obligation with a shape. Data you hold about buyers covers storage decisions made casually in month one that go wrong in year two, and the general principle is to hold what a possible dispute requires and nothing beyond it.

The monthly hour

One hour a month, on a date, doing the same four things.

Reconcile what arrived against what the platform says it sent. Timing differences and fee handling account for nearly every gap, and finding out which is exactly what reconciling against platform reports is for.

Update take-home per format and look at what it says rather than what you expected.

Check whether enquiries that did not convert are a traffic problem or a conversion problem, which is answered in one column - tracking the enquiries that did not convert.

And do it on a schedule rather than when you feel worried, because reviewing when worried guarantees you review at the least informative possible moment: reviewing your own numbers monthly makes the case better than a worried evening will.

What sits outside this

Three subjects are deliberately not covered here because they belong elsewhere. Measurement method and its conventions are documented properly on the site that owns them. The consumer-facing scoring tools your buyers have already used before they contacted you are described here rather than in a business post. And anything platform-specific - listing rules, payout methods, what a judge account actually offers - is the platform's own to publish, which for this one is Rate Cock's judges page.

The operating half of this trade is roughly two hours a month once it is running and roughly zero decisions, because they were all made once in advance. That is the whole trick, and it is why the earners still here in year two are usually not the ones who were best at the work in month one.